Prior distributions for our model are set based on previous literature and inputs from experts. Experts’ inputs are obtained via focused group discussion and survey.
The objectives of collecting prior information from literature:
The applicability and generalizability of data are determined by the extent of similarity between the study population and our population.
Category
Location
Population
Age
Education level
Applicability
Data
Taiwan
College students, Acedamic staff
18 to 60
Current students to Phd
NA
Match
Low
Medium
Medium
Medium
Low to Medium
Variable
Interest in learning (I)
Learning hours (T)
Teacher’s instructional attitude (Mo)
I*Mo
T*Mo
Direction and Degree of Association
Positive, Significant
Positive, Significant
Positive, Significant
Positive, Significantly strong
Positive, Significantly strong
Path Coefficient Value
0.462
0.451
0.442
0.691
0.672
When a teacher’s instructional attitude exerts a positive extraneous effect, H1, H2 and H3 appear relatively insignificant, compared with the increased significance of H4 and H5 statistically.
Category
Location
Population
Age
Education level
Applicability
Data
Indonesia
High school student
16 to 17
Grade 11
NA
Match
Medium
Medium
Medium
Medium
Medium
Variable
Teacher’s communication skills (CS)
Student’s motivation to learn (LM)
CS*LM
Direction and Degree of Association
Positive, Significant
Positive, Significant
Positive, Significant
R-square Value
NA
NA
0.216
Every 1 point increase in communication will be followed by an increase in learning outcomes of 0.155 and every 1 point increase in learning motivation will be followed by an increase in learning outcomes of 0.108.
Communication skills and learning motivation contribute to the value of student learning outcomes by 21.60%, while the remaining 78.40% is contributed by other factors not examined in this study. Other variables that were not examined were internal factors (health, intelligence, interests) and external factors (family and community environment).
Category
Location
Population
Age
Education level
Applicability
Data
Indonesia
High school student
16 to 17
Grade 11
NA
Match
Medium
Medium
Medium
Medium
Medium
Variable
Motivation (M) (self/external)
Habit (H)
M*H
Direction and Degree of Association
Positive, Significant, Weak
Positive, Significant, Weak
Positive, Significant, Weak
R-square Value
0.217
0.235
0.302
Category
Location
Population
Age
Education level
Applicability
Data
Indonesia
High school student
16 to 17
Grade 11
NA
Match
Medium
Medium
Medium
Medium
Medium
Variable
Creativity (X1)
Interest in Learning (X2)
X1*X2
Direction and Degree of Association
Positive, Significant
Positive, Significant
Positive, Significant, Strong
R-square Value
NA
NA
0.503
Student’s creativity, interest, and creativity and interest have an effect of 14%, 41.8% and 50.3% on student achievement in economics subjects for class XI Social Sciences SMA Ekasakti Padang, the remaining 49.7% is influenced by several other factors.
We chose 4 papers as reference for our initial priors. Our search focused on the papers that studied variables with potential impact on learning outcome in a traditional classroom setting. Intervention studies (novel teaching method, learning environment, hybrid classroom, etc) were excluded.
None of the selected studies has a population closely matched (high applicability) to our target population. Hence, the papers’ findings cannot be applied directly to our model, but they will be useful in building the basic framework of the model. Paper 1, 2, 3 and 4 quantified the relationship between variables (Changes in the Value of Outcome), which could be very useful in coding of our model. Unfortunately due to differences in population and study subject, we were not able to use those results.
Thus, we summarized the results- usable for our model- of the four papers as follows:
A conceptual model was drafted during focus group discussions with football players, coaches, cook/chef, sports scientist, medical and management staff. The conceptual model was drafted in three stages.
According to MFF management, the main goal or outcome of the intervention was to increase chances of winning the games. However, we need to consider if such outcome is achievable (measurable) with our intervention. To win a football match needs more than a good nutrition, and thus, good nutrition may not have a strong causal effect on “increased chance of winning a game.” Hence, we set the outcomes of our intervention as follows:
Note: The intervention for “Food quality and safety at kitchens” was already done during the 2-month internship.
This document focuses on the first outcome of our multistage model “the learning outcome”- improved nutritional knowledge and practice (eating habit). To achieve the second outcome “improved diet,” the MFF football players need to be provided with the right food in ample amount by the kitchens, on top of having good nutrition knowledge and eating habits. Hence, we did the food service intervention (Food quality and safety at kitchens) first. The conditions of food service after the intervention and chances of quality maintenance were considered in our model.
The focus group discussion for the first outcome was done with eight players from four teams- three men teams and one women team. They were grouped into two groups- one group for Intervention 1 and another for Intervention 2.
We explained in detail about the two interventions and provided some questions to spark the thinking process, before the groups were asked to draft model.
Two main questions were
What are the main contributing factors to meet each objective of the nutrition training?
(What do you need to learn and understand (apply) each lesson? e.g. visual aid, practice session)
What can hinder or support the main contributing factors?
(What can interfere with your learning? e.g. lack of energy after training)
The learning outcome was achieved 100 percent if all objectives of nutrition training were met.
Objectives of nutrition training for Intervention 1 (Nutritional self-reliance):
Objective of nutrition training for Intervention 2 (Guided personalized nutrition):
The players were first asked to list their answers to the questions (encouraged to add more points as needed) individually, and then discussed with their group members to create a new list. Based on the lists, model drafts were drawn by the two groups.
The groups presented their model, followed by an open discussion where each group provided additional inputs and suggestions for the other group. We collected the model drafts for intervention 1 and 2 after both groups approved the models. We then provided calibration training to the participants.
A survey questionnaire was developed based on the variables from the draft models and literature review model. A copy of questionnaire was given to the team captain of each team. The captain was asked to bring the questionnaire back to the team and fill in the estimates (lower and upper ranges) together with their team members. The forms were collected after a week.
Since we needed inputs from the players for all three models, we developed the questionnaire after drafting all models. Eight new players were requested for each modelling exercise. Thus, a total of 24 players (6 players from each team) were given calibration training after drafting the third and final model. The whole team- with the help of players trained in proper calibration- filled out the estimates for variables in survey questionnaire. After collecting all survey forms, we confirmed the core variables and drew the final multistage model based on the drafts.
Eight football players drafted the preliminary models for learning outcome.
Based on the two drafts, we designed our interventions.
Intervention 1: Nutritional Self-reliance
Intervention 2: Guided Personalized Nutrition
This was the last stage of “Priors for Learning Outcome.” We developed a survey form to assess the attitude, behavior and current nutrition knowledge of MFF players. The survey contained 20 questions that covered all three models.
As mentioned above, the form was given to the team captain after drafting all models, and the whole team would answer the survey together. We could use this approach because our population was small and homogeneous. The survey would allow us to estimate better ranges for input variables of our models.